Core Courses:

• Developed a visual regression model combining Transformer and CNN architectures using PyTorch to predict the body measurements of ducks.
• Developed data collection hardware and software facilities for depth and RGB cameras.
• Led a team to complete the collection of the dataset.
• Completed dataset analysis and cleaning using Python.
• Utilized the U-Net neural network for image segmentation as a preprocessing step.
• Completed the writing of the related research paper.

• Replaced the deep convolutional neural network in SRGAN’s discriminator with a Vision Transformer (ViT).
• Collected and augmented medical and satellite images for training purposes.
• Tuned relevant hyperparameters.
• Achieved superior SSIM and PSNR scores on multiple benchmark datasets compared to the unimproved SRGAN and other models.

• Trained and optimized a road sign recognition model using the latest YOLOv9.
• Adopted and optimized the TT100K dataset to implement data augmentation.
• Fine tuned relevant hyperparameters.

• Completed the distributed backend development using Spring Cloud.
• Implemented a distributed architecture using five databases and five microservices.
• Implemented service distribution using Spring Gateway.
• Implemented service registration and discovery for a cluster using Eureka.
• Constructed the database using MyBatis, MySQL, and Redis.
• Completed the front-end development using Vue.
• Awarded third prize in the National Competition of the Chinese Collegiate Programming Contest.

• Completed a comprehensive membership-based shopping mall using SpringBoot.
• Implemented database storage and interaction using MySQL and MyBatis.
• Completed the front-end development using Vue2.

• Positioned at the Intelligent Oil and Gas Center, Tsinghua University Research Institute in Shenzhen.
• Participated in well strata prediction for oil and gas fields based on machine learning.
• Utilized RobustScaler, smoothing, correction, sub-modeling, and knowledge-engineering-based data processing in the preprocessing and postprocessing stages, achieving a world-leading level in this field.
• Currently engaged in the core modeling and system development for predictive maintenance of mechanical equipment.
• Employed the TimesNet time series deep learning model, integrating supervised and semi-supervised learning models to construct an active learning framework. This framework is utilized for precise prediction of mechanical equipment failures under conditions of limited data availability.

• Collaborated with a professor to lead a student team of over ten members.
• Coordinated collaborations with institutions such as the Hunan Provincial Livestock and Veterinary Research Institute to conduct several provincial and national-level innovation projects.
• Responsible for the overall architecture design and development standards of front-end, back-end, and AI components for all laboratory projects.
• Responsible for the organization and preparation of multiple competitions.
• Participated as a principal in multiple research projects.